移动式激光雷达在人工林样地调查中的路径规划方法  

Path planning method of mobile lidar in plantation sample plot survey

作　　者：关天蒴 王春博 王蔚 闫飞[1,3] 范永祥 Guan Tianshuo;Wang Chunbo;Wang Wei;Yan Fei;Fan Yongxiang(Precision Forestry Key Laboratory of Beijing,Beijing Forestry University,Beijing 100083,China;Geophysical Survey Team of Hebei Province Coal Geological Bureau,Xingtai 054000,Hebei,China;Tibet Agricultural and Animal Husbandry University,Nyingchi 860000,Xizang,China;Jihua Laboratory,Foshan 528000,Guangdong,China)

机构地区：[1]北京林业大学精准林业北京市重点实验室,北京100083 [2]河北省煤田地质局物测地质队,河北邢台054000 [3]西藏农牧学院,西藏林芝860000 [4]季华实验室,广东佛山528000

出　　处：《北京林业大学学报》2024年第5期154-162,共9页Journal of Beijing Forestry University

基　　金：西藏自治区科技计划项目(XZ202301YD0043C);北京林业大学2021年大学生创新训练项目(S202110022009)。

摘　　要：【目的】林业调查规划对林业的可持续发展至关重要。在利用移动式激光雷达实现森林样地建图与量测过程中,全局一致性地图的准确性与扫描轨迹有着密切联系。因此,对样地观测路径进行合理规划尤为必要。【方法】本研究主要通过结合即时定位与地图创建(SLAM)技术,利用手持式激光雷达对样地进行扫描,根据SLAM技术应用的相关研究并结合林业样地特点规划出3种手持移动式激光雷达在样地内扫描路径方案并利用LeGO-LOAM算法实现点云地图构建,分析比较各路径建图效果及拟合量测的立木胸径、位置的精度差异。【结果】逐行式路径1拟合的样地立木胸径估计值偏差为2.18 cm,相对偏差为7.74%,均方根误差为2.74 cm,精度高于其他两种路径;在立木位置拟合精度方面,逐行式路径1与多环式路径3整体拟合效果较好,路径1精度略优于路径3,x轴估计值均方根误差(RMSE)为0.077 m,y轴估计值均方根误差在0.157 m,最大误差方向的协方差值为0.124 m,小于其他两路径。【结论】在使用32线激光雷达进行数据采集并基于LeGO-LOAM算法对森林样地进行点云建图进而实现单木因子进行拟合量测中,近似于航空摄影测量航线的路径方案的逐行式的路径1整体建图效果、量测精度相对优于其他路径,为地面移动式激光雷达外业数据采集提供一种合理的路径方案参考。[Objective]Forestry survey planning is important for the sustainable development of forestry.When using mobile lidar to do forest sample plot mapping and measurement,the accuracy of the global consistency map is closely related to the scanning trajectory.Therefore,the rational planning of sample plot observation path is of great significance.[Method]In this study,the forest sample plots were scanned by the handheld lidar with the use of simultaneous localization and mapping(SLAM).Based on the researches in the application of SLAM technology and the characteristics of forest sample land,three handheld mobile lidar scanning path plans in the sample plots were proposed.And the point cloud map was constructed through LeGO-LOAM algorithm.Thus,the three paths were compared and analyzed in terms of mapping effects,the accuracy of diameter of standing wood chest and the accuracy of its position(curve fitting data).[Result]The estimated diameter of the prototype stumpage chest diameter fitted by the progressive path 1 had a deviation of 2.18 cm,a relative deviation of 7.74%,and a root mean square error(RMSE)of 2.74 cm,and its accuracy was higher than the other two paths;path 1 fitted the rib diameter of the substation boundary standing wood and the internal standing wood better than path 2 and path 3;in terms of the standing wood position fitting accuracy,path 1 and path 3 had a better overall fitting effect,and the accuracy of path 1 was slightly better than path 3,with an x-axis estimated root mean square error(RMSE)of 0.077 m,a y-axis estimation root mean square error of 0.157 m,and a covariance difference of the maximum error direction of 0.124 m,all of which were smaller than the other two paths.[Conclusion]When collecting data through 32-line lidar,with the use of point cloud mapping of the forest-like land and the LeGO-LOAM algorithm,to achieve the fit measurement of the single wood factor,the overall mapping effect and the measurement accuracy of the aerial photogrammetry route like path plan(Path 1)are relatively

关 键 词：林业调查 激光雷达 SLAM技术 精度验证 

分 类 号：S758.7[农业科学—森林经理学]